Grazing by Marine Nanoflagellates on Viruses and Virus-Sized Particles: Ingestion and Digestion

l MARINE ECOLOGY PROGRESS SERIES Vol. 94: 1-10. 1993 Published March 3 1 Mar. Ecol. Prog. Ser. l

Grazing by marine nanoflagellates on viruses and virus-sized particles: ingestion and digestion

Juan M. Gonzalezl**,Curtis A. suttle2.**

'college of Oceanography, Oregon State University, Oceanography Admin. Bldg #104, Corvallis, Oregon 97331-5503, USA *~arineScience Institute, The University of Texas at Austin, PO Box 1267, Port Aransas,Texas 78373-1267, USA

ABSTRACT: We examined grazing of marine viruses and bacteria by natural assemblagesand cultures of phagotrophic nanoflagellates. Ingestion rates were determined using fluorescently labelled viruses (FLVs) and bacteria (FLB),and 50 or 500 nm diameter fluorescent microspheres (FMs). Calculated clearance rates of viruses by natural nanoflagellate assemblages were about 4 % of those for bacter~a when the bacteria and viruses were present at natural concentrations.Different viruses were ingested at different rates w~ththe smallest virus being ingested at the slowest rate. Further, we found differences in digestion times for the same flagellates grazing on different vlruses and for different flagellate assemblages grazing on the same viruses. FMs of 50 nm diameter were used as a control for egestion of undlgested particles. As rates of digestion were greater than those for ingestion both processes would occur simultaneously; hence, our estimates of grazing rate are likely conservative. Ingestion rates were positively correlated with the concentration of 50 nm FMs. Discriminat~onagalnst 50 nm FMs in favor of FLVs was also observed. Our calculations suggest that vlruses may be of nutritional sig- nificance for phagotrophic flagellates. When there are lob bacteria ml-l and 10' to 10' vvlruses ml-l, viruses may represent 0.2 to 9 % of the carbon, 0.3 to 14 '% of the nitrogen and 0.6 to 28 "/a of the phosphorus that the flagellates obtain from ingestion of bacteria. This study demonstratesthat both natural assemblages and cultures of phagotrophic nanoflagellates consume and digest a variety of marine vlruses, thereby deriving nutritional benefit and serving as a natural sink for marine viral particles. In addition, these results imply that some nanoflagellates are likely capable of consuming a wide spec- trum of organlc particles in the colloidal size range.

INTRODUCTION rioplankton assemblages contain viral particles, which implies that a significant fraction of bacterial and Although it is well established that viruses infect cyanobacterial production may be diverted into viral marine bacteria (e.g. Spencer 1955, Hidaka 1971, Moe- production (Bergh et al. 1989, Barrsheim et al. 1990, bus 1980), it was only relatively recently demonstrated Proctor & Fuhrman 1990, Heldal & Bratbak 1991). that the concentrations of virus-like particles in sea- Despite the great abundance of viruses in the sea water are typically in excess of 107 ml-', whether and turnover times estimated to range from hours to counted by electron or epifluorescent microscopy days (e.g. Berry & Noton 1976, Kapuscinski & Mitchell (Bergh et al. 1989, Proctor & Fuhrman 1990, Suttle et al. 1980, Heldal & Bratbak 1991, Suttle & Chen 1992) 1990, Hara et al. 1991, Paul et al. 1991). There are also much remains to be learned concerning the processes viruses which infect marine prokaryotic and eukaryotic responsible for the decay of infectivity and removal of phytoplankton (Mayer & Taylor 1979, Suttle et al. 1990, viral particles from seawater. Obviously a number of 1991). However, our understandmg of how viruses fit mechanisms potentially contribute to the decay of viral into aquatic foodwebs is still very incomplete. Estimates particles and infectivity in seawater including adhe- suggest that up to 16 % of the bacteria in natural bacte- sion to particulate material, bacterial exoenzymatic activity, chemical inactivation and degradation by solar radiation (e.g. Berry & Noton 1976, Kapuscinslu & ' Present address: Japan Marine Science and Technology Center (JAMSTEC),2-15 Natsushima-cho,Yokosuka 237, Mitchell 1980, Suttle & Chen 1992). Another possibility Japan is that viral particles are removed through grazing by ' ' Addressee for correspondenc~ phagotrophic flagellates.

0 Inter-Research 1993 2 Mar. Ecol. Prog. Ser. 94: 1-10, 1993

In this work, we report on a method for fluorescently decreased viral infectivity (data not shown). Following labelling viruses so that they are suitable for use as sonication and filtration, the FLVs were counted using tracers for the ingestion of viruses by phagotrophic epifluorescence microscopy (see below) and immedi- flagellates. Using this methodology and fluorescent ately inoculated into the water samples. The infectivity microspheres, we have investigated the potential of of FLVs following staining and sonication was tested isolates and natural assemblages of nanoflagellates to by plaque assays on the appropriate bacterial host. ingest marine bacteriophages and virus-sized parti- Electron microscopy. The viruses used for the graz- cles. We also examined digestion rates by quantifying ing studies were characterized morphologically using the disappearance of ingested viruses from within fla- electron microscopy. Samples either from amplified gellate food vacuoles. Our results confirm that being virus stocks or from freshly filtered fluorescently grazed by protists is one of the possible fates for labelled viral preparations were spotted onto 400 viruses in aquatic ecosystems. In addition, our calcula- mesh carbon-coated copper grids and allowed to tions indicate that viruses can contribute significantly adsorb for 30 min. The grids with the adsorbed to the nutrition of nanoflagellates. These results extend viruses were then rinsed through several drops of our concept of phagotrophic nanoflagellates as con- deionized-distilled water to remove salts and stained sumers of picoplanktonic cells, including virus-sized with 1 % w/v uranyl acetate and observed using particles as well. This further emphasizes the key role either a Joel JEM-1000X or Philips 301 transmission of flagellates in aquatic microbial foodwebs and sug- electron microscope. Procedures are outlined further gests that they may be even more important as re- in Suttle (1993). mineralizers than previously conceived. Ingestion rates. Aliquots (50 to 100 ml) of cultures or freshly collected seawater were poured into WhirlPak bags or polycarbonate flasks, which had been pre- MATERIALS AND METHODS soaked m 10 % (v/v) HC1, and rinsed with deionized water. To allow the protists to recover from handling Samples and enrichments. Seawater for grazing shock, experimental samples were incubated for experiments, enrichment cultures and isolation of fla- 30 min prior to the beginning of each experiment. gellates and viruses was collected from the pier at the Natural flagellate communities were from the Texas Marine Science Institute of The University of Texas at sampling location and were incubated at the in situ Austin (Port Aransas, Texas, USA) and from a sampling temperature (ca 30 'C). Cultures and enrichments of site located 5 km due west of Yaquina Bay (Oregon, flagellates were from Oregon and were incubated at USA). The bodonid isolate (E4,ca 5 X 8 pm in size) and 15 "C. Flagellate cultures were grown in 0.2 pm fil- the enrichments of natural flagellate communities orig- tered natural seawater plus 0.001 % yeast extract and inated with seawater collected from the Oregon sam- were used in grazing experiments during the late- pling site. Flagellate enrichments were prepared by exponential or stationary phases of growth. adding 0.001 % yeast extract (final concentration) to We compared the ingestion rates of flagellates on natural samples. Both monospecific cultures and nat- FLVs, 50 and 500 nm diameter fluorescent micro- ural enrichments were incubated in the dark at 15 "C spheres (FMs) (Polysciences, Inc., Warrington, Pa), and without shaking. The growth of associated bacteria fluorescently labelled bacteria (FLB) (Sherr et al. resulted in a yield of approximately 105 flagellates 1987). All treatments were duplicated. Prior to experi- ml-l. With the exception of bacteriophage T4, the ments the FMs were protein-coated in 5 mg ml-' albu- viruses used in these studies were isolated from Texas min solution for 24 h (Pace & Bailiff 1987). The FLVs coastal waters and were pathogens of marine bacteria. and 50 nm FMs (virus-sized particles) were added to Preparation of fluorescently labelled viruses the samples at a final concentration of about 107 ml-l, (FLVs). Viruses were fluorescently labelled by whereas the FLB and 500 nm FMs (bacteria-sized adding 0.5 p1 of a solution of 4 mg ml-' of DTAF (5- particles) were added at about 106 ml-l. Grazing rates [{4,6-dichlorotriazin-2-y1)aminojfluorescein)in 0.05M on different particles were determined in independent Na2HP0, to 1 m1 of viral suspension (ca 101° viruses), experiments. Approximately 44 to 53 % of the virus- mixing gently and incubating overnight at 4 "C in the sized particles and 3 to 33 % of the bacteria-sized par- dark. The DTAF solution was filtered through a 0.2pm ticles in the natural samples were comprised of the pore-size polycarbonate filter before use. Stained viral fluorescent surrogates. The effect of particle concen- suspensions were sonicated for 1 min in an ultrasonic tration on ingestion rates of natural flagellate assem- cleaner (Branson Ultrasonic Co.), and filtered through blages was corrected according to McManus & Okubo a 0.2 pm pore-size polycarbonate filter just prior to use. (1991). Sonication reduced clumping of the viruses; however, After the addition of the fluorescent particles, sam- longer sonication did not improve the results and ples were taken at 5 or l5 min intervals with the more Gonzalez & Suttle: Nanoflagellates grazing on viruses and virus-sized particles

frequent sampling being used for the experiments at contained the same concentration of bacteria as the ca 30 "C. The samples were fixed by the Lugol-Forma- original samples. The ingestion rates in the diluted lin decoloration technique (Sherr et al. 1988) to reduce samples were determined in controls in which the con- loss of material from the food vacuoles (Sherr et al. centration of fluorescent particles was the same as in 1989).The preserved samples were stained with 4',6- the experimental samples after dilution. Decreases of diamidino-2-phenylindole (DAPI) (Porter & Feig 1980) fluorescent particles within the protist cells after dilu- and filtered onto 0.8 pm pore-size polycarbonate fil- tion were used to calculate digestion rates. Digestion ters. A minimum of 30 phagotrophic nanoflagellates rates were calculated by regression analysis as previ- were inspected for each time period to determine the ously described (Sherr et al. 1988). The decrease in average number of fluorescent particles per cell. 50 nm FMs in the flagellates was used as a control for Sometimes, it was difficult to distinguish among 2 or the egestion of undigested virus-sized particles. Diges- more FLVs or 50 nm FMs contained in the same food tion times of FLV were estimated as the X-intercept of vacuole; consequently, they were counted as a single the digestion regression line. particle. This leads to conservative estimates of graz- We also compared flagellate ingestion and digestion ing rates. Stained viruses and 50 nm FMs were rates for T4 viruses stained with either DTAF or with counted directly on glass slides with an inverted epi- an FITC-labelled antibody. T4 viruses (Carolina Bio- fluorescence microscope at lOOOx (Suttle et al. 1991). logical Supply) were labelled with an anti-T4 antibody The FLB and 500 nm FMs were filtered onto 0.2 pm made in rabbit (Antibodies Incorporated) to which an pore-size polycarbonate filters and enumerated as anti-rabbit FITC-antibody (Sigma Co.) from goat was above. Bacteria were counted using the acridine conjugated. Immediately before use, labelled viruses orange direct count method (Hobbie et al. 1977).Rela- were 0.2 pm filtered to remove aggregates and possi- tive estimates of ingestion rates (fluorescent particles ble bacterial contamination. The grazing experiments cell-' min-l) and clearance rates (nl cell-' h-') were were conducted as outlined above. calculated from the uptake rates and concentrations of Statistical analysis. Statistical analyses were carried FLB in the experimental samples, as previously out according to Sokal & Rohlf (1981). A paired Stu- described (Fenchel 1980, Sherr et al. 1987).Flagellate dent's t-test was used to compare clearance and inges- grazing on different virus assemblages was compared tion rates of FLVs and FLBs by natural populations of by using the clearance rate data; ingestion rates were phagotrophic nanoflagellates. Regression and correla- compared to digestion rates of the same virus assem- tion analyses were used to relate ingestion rates and blage in each experiment. Absolute ingestion and densities of 50 and 500 nm FMs, and ingestion and clearance rates were calculated for natural assem- digestion rates of FLVs. Differences between slopes blages from estimates of relative grazing rate and were tested with the F-test for the difference between concentration of virus- and bacteria-sized particles 2 regression coefficients. Differences between clear- per unit volume. The rates were corrected for the ance rates of 50 nm FMs and FLVs, clearance rates of increased concentration of particles resulting from different viruses, and digestion times of different the use of surrogates to measure grazing rates (see viruses by different flagellate assemblages were car- McManus & Okubo 1991). ried out using analysis of variance (ANOVA).Planned We estimated the amount of carbon (C), nitrogen comparisons among the means were used for testing (N) and phosphorus (P) that natural assemblages of which means were significantly different from each phagotrophic nanoflagellates obtained from ingestion other. of viruses and bacteria using the data for absolute clearance rates. The C, N and P in bacteria and RESULTS viruses were assumed to be 2 X 10-l4 g C, 0.5 X 10-l4 g N, and 0.05 X 10-l4 g P per bacterium (Malone & Virus morphology Ducklow 1990), and 1 X 10-16 g C, 0.4 X 10-'" N, and 0.08 X 10-l6 g P per virus (Mathews et al. 1983, B0r- The marine bacteriophages used in the grazing sheim et al. 1990). experiments were characterized using electron Digestion rates. Digestion rate studies were carried microscopy, and micrographs of three of these (LMG1- out according to Sherr et al. (1988). Treatments and P4, PWH3a-P1 and LBIVL-Plb) are published else- controls were duplicated. The ingestion of fluores- where (Suttle & Chen 1992).LMG1-P4, PWH3a-P1 and cently labelled particles by flagellates in seawater LBlVM-Pla are of similar size and have head diame- samples or cultures was monitored. Once the average ters of approximately 78, 83 and 71 nm, and rigid tails number of particles per flagellate remained constant about 97, 104 and 86 nm in length, respectively. the cultures were diluted 10-fold with fluorescent- LBlVL-Plb is considerably smaller, with a head diam- particle-free, 0.2 pm filtered natural seawater which eter of about 50 nm and a very short tail of approxi- 4 Mar. Ecol. Prog. Ser. 94: 2-10, 1993

mately 11 nm. The LB viruses both infect a biolumines- were achieved when 0.5 p1 of the stock solution was cent bacterium that has tentatively been identified as added. Higher concentrations of stain resulted in a Photobacterium (Vibrio) leiognathi. The taxonomic background which made counting difficult. No fluo- status of the bacteria infected by the other phages is rescent particles were visible in the 0.2 pm filtered unknown. DTAF solution that could be confused with stained viruses. The FLVs were not washed after staining as this resulted in clumping of the particles. During the Fluorescently labelled viruses short duration of our experiments the particulate material in the samples was not noticeably stained by Several bacteriophages and an algal virus (data not DTAF that was introduced with the stained viruses. shown) were successfully stained using DTAF, and Staining the viruses at 4 "C was found to be optimum; even though most were

Ingestion experiments

We studied the ingestion of FLVs using natural populations, cultures and enrichments of phagotrophic nanoflagellates. In- gestion rates of FLVs and FLB by the flagellates were constant during the initial period of the incubations (Fig. 1). We also observed Time (minutes) that the relative ingestion rates (fluorescent particles cell-' min-') of FLVs were greater Fig. 1 Two representative ingestion (left) and d~gestion(right) experi- than those for FLB, when present at concen- ments using monospecific cultures of the bodonid E4. In (A), (0) T4 FLV. trations of about 106 and 107 ml-l, respec- (0) T4 labelled with FITC-conjugating antibody, and (A) 50 nm FMs were tively (Table 1). Because of the different compared. In (B),FLVs made from 2 marine virus isolates. (0) PWH3a-P1 and (0) LlMG1-P4, and (+) 50 nm FMs were compared. Error bars = SD of concentrations, however, when relative duplicate treatments clearance rates are compared (nl cell-' h-') Gonzalez & Suttle Nanoflagellates grazlng on vlruses and virus-sized particles 5

Table 1 Comparative grazing rates of fluorescently labelled vlruses (FLVs), fluorescently labelled bacteria (FLB), and 50 nm fluorescent microspheres (FMs) by natural populat~ons of phagotrophic nanoflagellates in waters from the Texas coast lndividual grazing rates were determ~nedin independent experiments. FPs: fluorescent viral-sized particles (FMs + FLVs) One SD of duplicate determinatlons IS given in parentheses

Flagellates ml' l FPs FLB lngestion rates Clearance rates Type x107 ml' x1O"ml' (fluorescent particles (nl cell" min-') cell- ' min. ') FPs FLB FPs FLB

1730 (340) 50 nm FMs 2.3 - 0.022 (0.002) - LBIVM-Pla 2.1 - 0.030 (0.005) - 380 (70) PWH3a-PI 1.1 0.4 0.054 (0.000) 0.030 (0.003) 860 (140) PWH3a-P1 1.1 0.9 0.031 (0.002) 0.028 (0.003) 890 (30) PWH3a-PI 1.6 2.1 0.043 (0.002) 0.048 (0.003)

those for FLBs were about 10-fold greater than those teria than on viruses in natural seawater samples for FLVs (p < 0.001). For natural assemblages of fla- (Table 1). gellates, absolute clearance rates on virus-sized parti- Clearance rates of 50 nm FMs were significantly lower cles ranged from 2.6 to 4.8 % of the rates on bacteria- (p < 0.001) than the corresponding clearance rates of sized particles (Table 1). FLVs by both natural populations (Table 1) and cultures lngestion rates by flagellates on 50 nm FMs were (Table 2) of phagotrophic nanoflagellates. This suggests strongly dependent on concentration (Fig. i),and discrimination against 50 nm FMs in favor of FLVs. We there was no evidence of saturation even at 10RFMs also observed significant differences in clearance rates ml-l. Thus, a one order of magnitude increase in the on different viruses. For instance, in both a bodonid cul- concentration of 50 nm FM5 resulted in a 45-fold ture and a flagellate enrichment, clearance rates on increase in ingestion rates by the protists. In contrast, PWH3a-P1 and LBlVL-Plb were lower (p < 0.05) than a similar increase in the concentration of 500 nm on LMG1-P4 (Table 2). Yet, ingestion rates were lowest FMs resulted in only a 5-fold increase in ingestion on the smallest virus (LBlVL-Plb).These results indi- rate (Fig. 2). A significant difference (p < 0.001) was cate that grazing rates on viruses will depend greatly found between the regression coefficients relating on both the virus and flagellate assemblages that are the concentrations of 50 and 500 nm FMs to inges- present. tion rates. Absolute clearance and inges- - tion rates (Table 1) were calcu- 7 lated using the regressions in G Fig. 2 to correct for the increased 7 particle concentrations resulting ; 10 -l! from the addition of surrogates

during the grazing experiments. In m our experiments carried out with 2 natural assemblages of flagellates (Table 1) comparing ingestion of viruses and bacteria, there were a total of 4.3 to 8.9 X 106 bacteria 10-3 ml-', 3 to 33 % of which were .: FLB, and 2.5 to 3.0 X 107 viruses ml-l, 44 to 53 % of which were E FLVs. Those calculations resulted IO-'~ - in estimates of flagellate clearance 10' lo6 10' 1 o8 10' rates that were 3 to 31 % lower for FM ml-I bacteria and 62 to 72 % lower for virus-sized *bsolute in- Fig. 2 Ingest~onrates of 50 nm (filled symbols) and 500 nrn (open symbols) diameter FMs as a functlon of FM concentration. Data are from experiments on natural gestion and 'learance rates were assemblages (squares) and cultures (circles). Regression lines are: log y = 14.001 + 3.6- to 13.7-fold and 20.8- to 38.5- 1.65610~~~(r = 0.969, n = 14, p < 0.001), for 50 nrn FMs, and log y = 7.560 + fold greater, respectively, on bac- 0.713 log X (r = 0.958, n = 6, p c 0.01) for 500 nm FMs 6 Mar. Ecol. Prog. Ser. 94: 1-10, 1993

Table 2. Results of some ingestion and digestion experiments comparing d~fferentFLV types and 50 nm FMs grazed upon by different flagellate assemblages. One SD in parentheses (n = 2)

Flagellates Viral-sized Conc. Ingestion rates Clearance rates Digestion rate" Digestion time particles (x107 ml-') (fluorescent particles (nl cell-' h-') (fluorescent particles (rnin) cell-' min.') cell-' min-l)

~odonid' LBlVM-Pla 2.0 0.034 (0.005) 0.102 (0.018) 0.059 (0.005) 60.0 (1.4) 50 nm FMs 2.8 0.023 (0.004) 0 049 (0.014)

BodonidC LMG1-P4 1.O PWH3a-P1 1.2 LBlVL-Plb 0.9 50 nm FMs 1.5 Flagellate LMG 1-P4 1.0 enrichmentr PWH3a-P1 1.2 LBIVL-Plb 0.9 50 nm FMs 1.6

Bodonld C T4 18 50 nm FMs 2.1

aDigestion rates are given as absolute values. 'Significant differences at the p < 0.001 level between ingestion and digestion rates '2.5 d old culture C6d old culture

Digestion experiments showed similar digestion times (non-significant differences) for 2 of the assayed FLVs (53 [SD = 0.81 rnin for Following the 10-fold dilution of the experimental LMG1-P4 and 52 [SD = 3.21 rnin for PWH3a-PI), but samples with FLV-free seawater, the number of in- LBlVL-Plb (46 [SD = 0.61 min) was digested faster (p < gested FLVs per flagellate decreased linearly with time 0.01). However, a flagellate enrichment showed the (Fig. 1).In contrast, the concentration of ingested 50 nm following digestion times: 37 (SD = 0.8), 52 (SD = 0.7), FMs remained constant for the first 30 rnin subsequent and 47 (SD = 1.4) rnin for LMG1-P4, PWH3a-PI, and to dilution. LBlVL-Plb, respectively, which represents significant Significant differences in digestion times were differences (p < 0.001) among them. We observed that observed among different flagellates grazing on the a bodonid culture and a flagellate enrichment had sim- same viruses and among the same flagellates grazing ilar digestion times (non-significant differences) for the on different viruses (Table 2). A bodonid culture viral strain PWH3a-P4 and for LBlVL-Plb. Neverthe- less, LMG1-P4 was digested more rapidly (p < 0.001) by the flagellate enrichment than by the bodonid culture. A comparison of ingestion and digestion rates of FLVs (Table 2) indicated that digestion rates were significantly (p < 0.001) faster than ingestion rates, although the rates were correlated with each other (r = 0.985, n = 8, p < 0.001) (Fig. 3).

DISCUSSION

0.01 0.02 0.03 0.04 A number of important results emerged from this Ingestion rate study. First, we were able to modify an existing technique to fluorescently label marine viruses so that they could be used as tracers of natural marine virus com- Fig. 3. Relationship between ingestion and digestion rates of munities. Second, we demonstrated that viruses were FLVs by different flagellate cultures. Ingestion and digestion rates are expressed in FLVs cell-' min-'. Regression line is ingested and digested by natural assemblages and cul- y = 0.011 t 1.163~(r = 0.985, n = 8, p c 0.001) tures of manne nanoflagellates. Third, we showed that Gonzalez & Suttle: Nanoflagellates gr .azing on viruses and virus-sized particles 7

ingestion and digestion rates depended on the virus natural populations of phagotrophic nanoflagellates at being grazed and the flagellate grazer. These results rates that are similar to those for bacteria, when both are discussed in detail below. bacteria and viruses are present at natural concentrations. The ingestion rates that we observed for PWH3a- P1 (Table 1) ranged from 1.9 to 3.2 viruses cell-' h-' Fluorescently labelled viruses when the viruses were present at 1.1 to 1.6 X 107ml' This is very similar to reported ingestion rates of In this study we prepared fluorescently labelled PWH3a-P1 (3.3 viruses cell-' h-') based on the decay viruses using a stain (DTAF) which had been of infectious viruses in the presence of heterotrophic en~ployedto stain bacteria (Sherr et al. 1987) and nanoflagellates (Suttle & Chen 1992). In addition, phytoplankton (Rublee & Gallegos 1989, Sherr et al. although flagellates are usually selective for larger 1991). Using this method we stained several marine particles (Gonzalez et al. 1990b),there may be compo- bacteriophages and an algal virus, which subse- nents of the flagellate community that are specialist quently were visible by epifluorescence microscopy. grazers on viruses and virus-sized particles. For exam- The method is probably suitable for staining a wide ple, certain marine choanoflagellates in nature have variety of viruses. During the staining procedure it is been observed to restrict their grazing to virus-sized important to prevent the viruses from aggregating as particles (J. M. Gonzalez, C. A. Suttle. E. B. Sherr & they are difficult to disperse. We accomplished this by B. F. Sherr unpubl.). minimizing the handling of the viruses, staining at Interestingly, flagellates ingested different viruses at 4 "C, sonicating for 1 min, and then filtering the solu- different rates, implying that selective grazing was tion through 0.2 pm pore size polycarbonate filters. occurring although we do not know the basis of this Filtration also removed any contaminating bacteria selection. However, a natural flagellate assemblage from the FLV suspension. Using transmission electron and a bodonid culture ingested the smallest virus at microscopy we found that viruses prepared in this the slowest rate. As viruses vary considerably in size, manner were present essentially as individual free- shape, morphology (e.g. tail structure), and surface viral particles; however, we recommend that investi- charge there are a number of parameters that are gators check their preparation procedure by electron likely important in determining ingestion rates. microscopy, as well. DTAF-stained viruses were found Comparisons between the disappearance of FLVs to be suitable for estimating protozoan grazing rates and FMs from flagellate food vacuoles, subsequent to on viruses and potentially could be used for other ap- dilution with fluorescent-particle-free seawater, sug- plications where fluorescently labelled viruses would gest that the viruses were digested. Dubowsky (1974) be useful as tracers. In contrast, viruses labelled by has shown that disappearance of FMs from within FITC, conjugated to an antibody, were found to be flagellate food vacuoles is the result of egestion. Also, unsuitable for tracing virus ingestion by flagellates. observations of partially digested viruses inside the food vacuoles of flagellates (J. M. Gonzalez, C. A. Sut- tle, E. B. Sherr & B. F. Sherr unpubl.) provides convin- Ingestion and digestion rates of viruses cing evidence that the viruses are digested although the possibility that the DTAF stain disappears more Estimates of relative clearance rates for flagellates rapidly than the viruses are digested, cannot be dis- grazing on FLB were about 10-fold higher than those counted. Furthermore, egestion of intact or partially on FLVs. Similar differences in clearance rates have digested viruses is possible, as the process is thought to been found between 50 nm FMs and 500 nm FMs for occur when some organisms graze on bacteria (Taylor other natural flagellate assemblages (J. M. Gonzalez, & Berger 1976, King et al. 1988, Sherr et al. 1988, C. A. Suttle, E. B. Sherr & B. F. Sherr unpubl.). In Gonzalez et al. 1990a). nature, viral and bacterial abundances typically differ Our results indicate that viruses were digested more by a factor of about 10 (Bergh et al. 1989, Bratbak et al. rapidly than they were ingested (Table 2). Moreover, 1990, Proctor & Fuhrman 1990, Paul et al. 1991) digestion times varied among different viruses grazed although differences as large as 1000-fold have been by the same flagellate assemblage, and among differ- reported (Proctor & Fuhrman 1990). Therefore, ent flagellate assemblages grazing on the same although clearance rates (nl cell-' h-') are higher on viruses. Similar results have been reported for flagel- bacteria-sized than on virus-sized particles, ingestion lates grazing on bacteria (Sherr et al. 1983, Mitchell et rates (fluorescent particles cell-' min-') could be simi- al. 1988, Gonzalez et al. 1990a). lar or even greater for virus-sized particles under cer- The ingestion rates that we report for viruses may tain circun~stances.Nonetheless, our results conclu- be underestimated because of the conservative sively demonstrate that viruses can be ingested by approaches that we employed in counting fluorescent Mar. Ecol. Prog. Ser. 94: 1-10, 1993

particles within food vacuoles (see 'Materials and increase. When the relative concentrations of viruses methods') and in estimating grazing rates (see and bacteria differ by 5-fold (i.e. 5 X 106 viruses and 'Results'), and because of digestion of the viruses dur- 106 bacteria ml-l) viruses would constitute 0.1, 0.2 and

ing the period over which ingestion rates were deter- 0.3 O/oof the C, N, and P contributed by bacteria to the mined. Therefore, the importance of viruses as a nutri- flagellate diet. When the relative concentrations differ tional source for flagellates may be greater than by 50-fold (i.e. 5 X 107 viruses and 106 bacteria ml-l) indicated here. For instance, PWH3a-PI, the viral the relative contribution by viruses would be 1.0, 1.5 strain used for comparing clearance rates on viruses and 3.1 %. A 500-fold difference in the relative con- and bacteria by natural assemblages of nanoflagel- centration of bacteria and viruses (i.e. 5 X 107 viruses lates, showed a lower clearance rate than other viruses and 105 bacteria ml-l) would result in viruses con- tested (Table 2). Hence, clearance rates on other tributing 9.6, 15.4 and 30.7 %, respectively, of the C, N, viruses might provide estimates much higher (up to and P supplied by bacteria. These calculations indicate 100 %) than those reported. Furthermore, several that viruses can be a significant source of nutrients to authors (Muller et al. 1965, Stolze et al. 1969, Wetzel & nanoflagellates when viruses are present at concentra- Korn 1969, Dubowsky 1974) have shown that the tions greater than 50 times that of bacteria. Similar rel- digestive system in a variety of protozoa is activated ative concentrations of viruses and bacteria have been upon the formation of particle-containing vacuoles. reported for several aquatic ecosystems (Bergh et al. The results we obtained using viruses that were 1989, B~rsheimet al. 1990, Proctor & Fuhrman 1990, labelled with an FITC-conjugated antibody also sug- Heldal & Bratbak 1991). gest that digestion of food particles is rapidly initiated. Ingestion rates of flagellates have been shown to be If ingestion rates are corrected for digestion using the related to the number of prey available and typically data in Fig. 3 then estimates of grazing rates on viruses the rates saturate at high prey densities. For example, by natural assemblages of Ilagellates are up to 34 ?/0 ingestion rates on bacterial-sized particles saturate at higher than those reported in Table 1. concentrations of about 107 bacteria ml-' (Fenchel Although the flagellates were able to graze 50 nm 1982, Rassoulzadegan & Sheldon 1986). Yet, we found FMs, the clearance rates we obtained were lower than no evidence of saturation at densities of virus-sized those measured using FLVs (Table 1 & 2). Similar FMs up to 10' ml-' (Fig. 2). Moreover, ingestion rates results have been reported for bacterial-sized micro- on virus-sized particles were strongly dependent on spheres (Pace & Bailiff 1987, Sherr et al. 1987) concentration; a 10-fold increase in concentration although some protists do not show significant differ- (i.e. from 107 to 108 ml-l) resulted in approximately a ences between ingestion of FMs and FLB (Sherr et al. 45-fold increase in ingestion rate (Fig. 2). In contrast, a 1987, Sanders et al. 1.989). Our results of ingestion 10-fold increase in the concentration of bacterial-sized rates on 500 nm FMs and FLB are in agreement with particles (i.e. from 105 to 106 ml-') resulted in only reported ingestion rates on FMs (Pace & Bailiff 1987, about a 5-fold increase in ingestion rate. Hence, the Sherr et al. 1987) and FLB (Sherr et al. 1987, 1989), contribution of viruses to the nutrition of nanoflagel- respectively, by heterotrophic nanoflagellates. There- lates is proportionally much greater at high viral densi- fore, one must be cautious if 50 nm FMs are used as a ties. For example, when there are about 108 viruses surrogate for viruses in grazing experiments. and 10"acteria ml-l (Bergh et al. 1989, Bratbak et al. 1990, Proctor & Fuhrman 1990, Heldal & Bratbak 1991), viruses could supply phagotrophic nanoflagel- Ecological implications lates with a minimum of 9, 14 and 28 % of the C, N and P that they receive from ingestion of bacteria. We estimated the relative contributions of viruses Results from this study suggest that phagotrophy by and bacteria to the C, N and P nutrition of flagellates nanoflagellates is of limited importance as a loss over the range of relative densities of viruses and bac- process for natural virioplankton communities. Our teria reported in the literature (Bergh et al. 1989, B0r- data (Table 1) would imply turnover times of virus sheim et al. 1990, Bratbak et al. 1990, Proctor & communities on the order of years if grazing by Fuhrman 1990. Heldal & Bratbak 1991, Paul et al. nanoflagellates was the only loss process responsible 1991). These calculations were made using the aver- for the removal of viruses. age clearance rates for nanoflagellates grazing on bac- Grazing by nanoflagellates is another mechanism teria or viruses (Table l), and assuming that these rates besides infection which incorporates viruses into the C, were constant. This is a conservative assumption as the N, and P cycles of aquatic systems. Our results, cou- data in Fig. 2 suggest that the relative difference pled with observat~onsthat nanoflagellates can ingest between the clearance rates on bacteria- and virus- high-molecular-weight dissolved organic matter sized particles increases as the concentrations of both (Sherr 1988), also suggest that the large pools of sub- Gonzalez & Suttle: Nanoflagellates grazing on ~II-usesand virus-sized particles

micron-sized particles which are present in seawater Kapuscinski, R. B., Mitchell. R. (1980). Processes con- (Koike et al. 1990, Wells & Goldberg 1991) may be trolling virus ~nactivationin coastal waters. Wat. Res. 14: accessible to grazing by flagellates. Clearly, current 363-371 King. C. H.. Shotts, E. B. Jr. Wooley, K.E., Porter, K. G. (1988). concepts of microbial processes in the sea must be Survival of coliforms and bacterial pathogens within pro- altered to include grazing of viruses and virus-sized tozoa during chlorination. Appl. environ. Microbiol. 54: particles by flagellates. As well, our study reinforces 3023-3033 the paradigm that phagotrophic nanoflagellates are Koike. I., Hara, S., Terauchi, K., Kogure, K. (1990). Role of sub-micrometer particles In the ocean. Nature 345: key elements of nutrient cycles in marine ecosystems. 242-244 Malone, T. C., Ducklow, H. W. (1990). Microbial biomass in Acknowledgements. We appreciate the support and helpful the coastal plume of Chesapeake Bay: phytoplankton- comments by Drs Evelyn and Barry Sherr, the insightful dis- bacter~oplankton relationships. Limnol. Oceanogr 35: cussions with Dr S Stro~nand the technical assistance of 296-312 A. M. Chan and F. Chen. We are grateful for the constructive Mathews, C. K.,Kutter, E. M,,Mosig, G., Berget, P. B (eds.) comments of the reviewers. This research was supported by (l983). Bacteriophage T4. American Society for M~crobiol- grants OCE-9018833 (NSF) and N00014-90-5-1280 (ONR) to ogy, Washington, DC C.A.S.,OCE-8816428 (NSF) and OCE-8823091(NSF) to Eve- Mayer, J A., Taylor, F. J. R. (1979). A virus which lyses the lyn and Barry Sherr, and a postdoctoral fellowship from the marine nanoflagellate Micrornonas pusilla. Nature 281: Spanish Ministry of Education and Science to J.M.G. Contri- 299-301 bution no. 845 of the Marine Science Institute, The University McManus, G. B., Okubo, A. (1991). On the use of surrogate of Texas at Austin. food particles to measure protistan ingestion. Limnol. Oceanogr. 36: 613-617 Mitchell, G. C., Baker, J. H., Sleigh, H. A. (1988).Feeding of a LITERATURE CITED freshwater flagellate, Bodo saltans, on diverse bacteria. J. Protozool. 35: 219-222 Bergh, O., Bsrsheim, K. Y., Bratbak, G., Heldal, M. (1989). Moebus. K. (1980). A method for the detection of bacterio- High abundance of viruses found in aquatic environ- phages from ocean water Helgolander Meeresunters. 34: ments. Nature 340: 467-468 1-14 Berry, S. A., Noton, B. G. (1976).Survival of bacteriophages in Muller, M., Rohlich, P., Toro, I. (1965).Studies on feeding and seawater. Wat Res. 10: 323-327 digestion in protozoa. VII. Ingestion of polystyrene latex Bsrsheim, Y.,Bratbak, G., Heldal, H. (1990).Enumeration and particles and its early effect on acid phosphatase In Para- biomass estimation of planktonic bacteria and viruses by meciuni multinucleaturn and Tetrahyniena pyriformis. transmission electron microscopy. Appl. environ. Micro- J Protozool. 12: 27-34 biol. 56: 352-366 Pace, h4. L., Bailiff, M. D. (1987). Evaluation of a fluorescent Bratbak, G., Heldal, M., Norland, S., Thingstad, T F. (1990). mlcrosphere technique for measuring grazlng rates of Viruses as partners in spring bloom microbial trophody- phagotrophic microorganisms. Mar. Ecol. Prog. Ser. 40: namics. Appl. environ. Microbiol. 56: 1400-1405 185-193 Dubowsky, N. (1974). Selectivity of ingestion and digestion in Paul. J. H.. Jiang, S. C., Rose, J. B. (1991). Concentration of the chrysomonad flagellate Ochromonas malharnensis. viruses and dissolved DNA from aquatic environments by J. Protozool. 21: 295-298 vortex flow filtration. Appl. environ. Microbiol. 57: Fenchel, T (1980). Suspension feeding in ciliated protozoa: 2197-2204 functional response and particle size selection. Microb. Porter, K. G., Feig, Y S. (1980). The use of DAPI for identify- Ecol. 6: 1-11 ing and counting aquatic microflora. Limnol. Oceanogr. Fenchel, T (1982). Ecology of heterotrophic microflagellates. 25: 943-948 11. Bioenergetics and growth. Mar. Ecol. Prog. Ser. 8: Proctor, L. M., Fuhrman, J. A. (1990). Viral mortality of marme 225-231 bacteria and cyanobacteria. Nature 343: 60-62 Gonzalez, J. M.. Iriberri. J., Egea, L.,Barcina, 1. (1990a). Dif- Rassoulzadegan, F., Sheldon, K. W. (1986). Predator-prey ferential rates of digestion of bacteria by freshwater and interactions of nanozooplankton and bacter~ain an olig- marine phagotrophic protozoa. Appl. environ. Microbiol. otrophic marine environment. Limnol. Oceanogr 3 1. 56: 1851-1857 1010-1021 Gonzalez, J. M., Sherr, E. B, Sherr, B. F. (1990b). Size-selec- Rublee, P. A., Gallegos, C. L. (1989). Use of fluorescently tive grazing on bacterla by natural assemblages of estuar- labeled algae (FLA) to estimate microzooplankton graz- ine flagellates and ciliates. Appl. environ. Microbiol. 56: lng. Mar. Ecol. Prog. Ser. 51: 221-227 583-589 Sanders, R. W., Porter, K. G., Bennett. S. J., DeBiase, Hara, S., Terauchi, K.. Koike. I. (1991). Abundance of viruses A. E. (1989). Seasonal patterns of bacterivory by in marine waters: assessment by epifluorescence and flagellates, ciliates, rotifers, and cladocerans in a fresh- transmission electron microscopy. Appl. environ. Micro- water planktonic community. Limnol. Oceanogr. 34: biol. 57: 2731-2734 673-687 Heldal, M,. Bratbak. G. (1991). Production and decay of Sherr, B. F., Sherr, E. B., Berman, T (1983). Grazing, growth viruses in aquatic environments. Mar. Ecol. Prog. Ser. 72: and ammonium excretion rates of a heterotrophic 205-212 microflagellate fed with four species of bacteria. Appl. Hidaka, T. (1971).Isolation of marine bacteriophages from sea environ. Microbiol. 45: 1196-1201 water. Bull. Jap. Soc. Scient. Fish. 37: 1199-1206 Sherr, B. F., Sherr. E. B.. Fallon, R. D. (1987). Use of mono- Hobbie, J. E.. Daley, R. J., Jasper, S. (1977). Use of Nuclepore dispersed, fluorescently labelled bacteria to estimate in filters for counting bacteria by fluorescence microscopy. situ protozoan bacterivory. Appl. environ. Microbiol. 53: Appl. environ. Microbiol. 33: 1225-1228 958-965 Mar. Ecol. Prog. Ser. 94: 1-10. 1993

Sherr, B. F., Sherr, E. B.,Rassoulzadegan, F. (1988). Rates of 396-409 digestion of bacterla by marlne phagotrophic protozoa: Suttle, C. A. (1993). Enumeration and isolat~onof viruses. In: temperature dependence. Appl. environ. Microbiol. 54: Kemp, P. F.. Sherr, B. F., Sherr, E. B., Cole, J. J. (eds.) 1091-1095 Current methods in aquatic microbiology Lewis Publ., Sherr, E. B. (1988).Direct use of high molecular weight poly- Chelsea, M1 (in press) saccharide by heterotrophic flagellates. Nature 335: Suttle, C. A., Chan, A. M., Cottrell, M. T (1990). Infection of 348-351 phytoplankton by viruses and reduction of primary pro- Sherr, E. B., Rassoulzadegan, F, Sherr, B. F. (1989). Bac- ductivlty. Nature 347: 467-469 terivory by pelagic choreotrichous ciliates in coastal Suttle, C. A., Chan, A. M., Cottrell, M. T (1991). Use of ultra- waters of the NW Mediterranean Sea. Mar. Ecol. Prog. filtration to isolate viruses from seawater which are Ser 55: 235-240 pathogens of marine phytoplankton. Appl. environ. Sherr, E. B., Sherr, B. F., McDaniel, J. (1991). Clearance rates Microbiol. 57. 721-726 of <6 pm fluorescently labeled algae (FLA) by estuarine Suttle, C. A., Chen, F. (1992). Mechanisms and rates of decay protozoa: potential grazing impact of flagellates and cili- of marine viruses in seawater. Appl. environ. Microbiol. ates. Mar. Ecol. Prog. Ser. 69: 81-92 58: 3721-3729 Sokal, R. R.,Rohlf, F. J. (1981). Biometry, 2nd edn. W. H. Free- Taylor, W. D., Berger, J. (1976). Growth responses of cohabit- man and Co., New York ing ciliate protozoa to various prey bacteria. Can. J. Zool. Spencer, R. (1955). A marine bacteriophage. Nature 175: 54: 1111-1114 690 Wells, M. L., Goldberg, E. D. (1991). Occurrence of small col- Stolze, H. J., Lui, N. S. T., Anderson, 0. R., Roels, 0.A. (1969). loids in seawater. Nature 353: 342-344 The influence of the mode of nutrition on the digestive Wetzel, M. G., Korn, E. (1969). Phagocytosis of latex beads by system of Ochromonas malhamensis. J Cell Biol. 43: Acanthamoeba castellanii (Neff). J. Cell Biol. 43: 90-104

Thisarticle was presented by D. A. Caron, Woods Hole, Manuscript first received: May 29, 1992 ~Massachusetts, USA Revised version accepted: December 2, 1992